‘It Must Have Been Stolen’

How can you spot a stolen election? Maybe just as important, how can you spot an election that isn’t stolen? You can never be entirely certain when the margins are fairly close, which is why stolen-election theories hold a natural attraction for conspiracy-minded partisans. But in terms of evidence, there are three signs to look for that might show that an election’s outcome was the result of fraud by voters or election officials: (1) direct evidence of illegally counted or discounted votes, (2) evidence of an unlawful process, and (3) anomalous results that make sense only if the election was fraudulent.

In the 2020 presidential election, Donald Trump and his supporters have claimed to find all three. But the evi-dence presented of actual fraud has been persistently underwhelming and often just wrong. Trump’s legal team has, in multiple cases, withdrawn claims that it could prove fraud when it came time to lay its proof on the table. Everywhere it has tried to prove fraud with evidence in court, it has failed. Even the best argu-ments for direct proof of fraudulent votes end up a long distance from the tens of thousands of votes across multiple states that would be needed to overturn the outcome. Multiple conser-va-tive judges, several of them Trump appointees to the federal bench, have reached this conclusion.

Evidence of procedural irregularities has likewise come up short. Un-question-ably, some rules were bent or mutilated in ways that favor Democrats, ranging from the Pennsylvania supreme court’s throwing out the state’s mail-in ballot deadline to a lenient approach taken in various states to curing defective mailed ballots (including fixing mistakes such as a missing signature). But the deadline extensions did not affect enough votes to change the outcome in any state; neither did any other legal issues or skullduggery in the obstruction of observers. Baroque theories of computerized vote-counting fraud have proven even more fantastical. It turns out to be untrue that Dominion Voting Systems used the Smartmatic software blamed for Hugo Chávez’s manipu-lation of Venezuelan elections; it also turns out to be untrue that Dominion has any ties to Democrats aside from having donated some voting machines to the Clinton Foundation for use in the third world. Many other things alleged have proven untrue, and the Trump legal team has not inspired confidence with its many rudimentary errors, such as challenging the results of precincts in Michigan that turned out to be in Minnesota.

Given the absence of evidence of election fraud of a size sufficient to alter the result, Trump supporters who believe the election was stolen have fallen back on the most amorphous category: anomalous results. This election, we are told, doesn’t make sense. The problem is, it does.

The first argument is that it does not make sense that Joe Biden, a 78-year-old political fossil who has been running for president and failing for 33 years, and who spent the bulk of the campaign season hiding in his basement and tripping over his sentences, could have inspired 81 million Americans to vote for him against an incumbent president who held big rallies and exudes vigor and brio. Trump supporters may find this dispiriting, but it does not require a great flight of imagi-nation to see how it could happen. Biden was on two national tickets that commanded a majority of the national popular vote, something Trump never did. Biden swept away a field of challengers more inspiring to Democratic partisans and activists, almost entirely on the basis of voters who wanted to ensure a candi-date who would help them beat Trump. He was not loved, but people turned out for him anyway.

Sure, it is easy to look at Biden and ask, “How could we possibly lose to this guy?” But Democrats are at least equally baffled that 63 million Americans voted for Donald Trump in 2016 and, after four years of watching him in office, that 74 million did in 2020. The candidates on offer in both 2016 and 2020 were deeply distressing to a lot of Americans, many of whom no longer understand their neigh-bors and most of whom decided to choose what they saw as a lesser evil. Trump, in particular, spent four years inflaming his critics’ loathing of him. He made the infuriating of liberals (“owning the libs,” in Internet-speak) central to his brand. Should we be surprised that liberals turned out in droves, if not to support Biden, then simply to stop being infuriated by Trump?

Yes, Biden held few in-person events, and drew far fewer in-person votes. But Bi-den’s supporters were disproportionately people who preferred to err on the side of caution. For months, Democrats preached that in-person voting was unsafe; for months, Republicans preached that mail-in voting was untrustworthy. It should sur-prise nobody that the two parties’ voters behaved in starkly different fashion.

As for the late-night “vote dumps” in Wisconsin, Michigan, and Pennsylvania, all of them resulted from mail-in ballots being counted late. Republicans have long, dark, ancestral memories of Democratic precincts voting late, with the specter of some Mayor Daley figure asking, “How many does Kennedy need?” This goes back to 1916, when Charles Evans Hughes went to bed thinking he had won Cali-fornia, and 1948, when Tom Dewey was prematurely declared the winner by a Chicago newspaper. But the timeline of vote counts was so predictable in 2020 that it had a name before Election Day: the “red mirage.” Because Democrats were more likely to vote by mail, and because the most heavily Democratic cities already tend to be the last counters owing to urban inefficiency, it was widely predicted that in those cities the counting of mail-in ballots would delay the most Democratic portion of the vote tally until the end. This did not happen everywhere: States such as Florida and Texas allowed mail-in ballots to be tabulated before Election Day. But Republican legislatures in the Midwest blocked early counting, and the result was in fact a high concen-tration of Democratic ballots at the end. Everybody who was paying attention saw this coming a mile away.

The Texas v. Pennsylvania lawsuit that was dismissed out of hand by a unanimous Supreme Court included an “expert” analysis claiming that there was a “one in a quadrillion” chance that these late ballots would break so heavily Democratic. As Robert VerBruggen observed on National Review Online, theories such as that one “simply assume that different batches of ballots should have similar breakdowns by candidate or party.” Thus “if Biden got more support than Clinton had, or if late-counted ballots were more heavily Biden-leaning than early ballots, that’s treated as evidence of fraud.” Such theories ignore three obvious realities. One, the 2016 election was a different setting from 2020, in that the Democrats had been in power for eight years. Two, Biden was a less disliked candidate than Hillary Clinton. Three, the patterns of mail-in voting were affected by the pandemic. The underlying logical flaw is the assumption that voters and their opinions do not change from one election to the next. The whole history of American democracy argues otherwise.

For that matter, in terms of probabilities, the margins of both the 2016 and 2020 elections — Trump winning three states by less than a point, Trump losing three states by less than a point — amount to little more than a coin toss. A coin landing heads-up three times in a row followed by tails-up three times in a row is mildly improbable; it does not mean the coin is loaded.

Is it shocking that Biden added almost 19 million votes to the Democrats’ column in four years? Again, historically, no. Turnout overall was up by 15.6 percent, which is lower than the 16 percent increase from 2000 to 2004, let alone the 26.5 percent increase from 1924 to 1928 or the 26.6 percent increase from 1948 to 1952. Biden increased his party’s vote totals in one cycle by less than Bush did in both 2000 and 2004, Ronald Reagan in 1984, Jimmy Carter in 1976, Richard Nixon in 1972, Lyndon Johnson in 1964, Dwight Eisenhower in 1952, John F. Kennedy in 1960, Wendell Willkie in 1940, Franklin Roosevelt in both 1932 and 1936, and Herbert Hoover and Al Smith in 1928.

Some have claimed that Biden did suspiciously well in four key swing-state cities: Philadelphia, Detroit, Milwaukee, and Atlanta. We are told that Biden did worse than Hillary Clinton in every major metro area except these four. This is almost the opposite of the truth. Look at the 50 largest cities in the country. If you count New York City’s five counties as one, these cities are based in 48 urban counties (Long Beach, Calif., shares a county with Los Angeles, and Mesa, Ariz., shares a county with Phoenix). Clinton won 41 of the 48 counties; Biden won 45. In 44 of those 48 counties, Biden’s margin of victory was greater than Clinton’s (or his margin of defeat was smaller). More-over, Philadelphia was one of the four where he did not win by more votes than Clinton had. In fractional terms, 36 out of 48 swung in Biden’s direction. And in Wayne County, Mich. (home of Detroit), Trump was the first Republican since 1988 to crack 30 percent of the vote.

The numerical reality, when compared with that of 2016, is that Trump did not lose the election in the big, blue, machine-run Democratic cities. He lost it in the suburbs of Atlanta, Phoenix, Philadelphia, and Detroit, and in and around Madison, Wis. But neatly manicured suburbs do not make for a convincing narrative of fraud.

Another supposedly anomalous result is the contrast between Trump and down-ticket Republicans. How, we are asked, could Republicans gain at least ten seats in the House while losing the White House? But this has happened before. In 1892, in a close general election coming off a midterm wipeout, Republicans regained 38 House seats while the Republican in-cumbent lost. In 1916, Republicans picked up 19 seats while failing to unseat the Democratic incumbent. In 1960, Nixon lost to Kennedy while House Republicans gained 22 seats. Moreover, nationally, the Democrats won the popular vote in House races by a margin, 50.8 percent to 47.7 percent, that does not look very different from that of the presidential race.

Some Senate Republicans ran far ahead of Trump, but this, too, is historically normal. In Maine, Susan Collins got 70,000 more votes in her race than Trump did in his; in 2008, she ran 150,000 votes ahead of presidential candidate John McCain. Presidential candidates often diverge from their party in the Senate. State by state, McCain ran, on average, four points ahead of Republican Senate candidates; in 1980, on average, Jimmy Carter ran almost ten points behind Democratic Senate candidates.

Ticket-splitting has been a feature of American politics forever. Skeptics argue that Republican voters would not choose Republican candidates for House and Sen-ate without also supporting Trump. They point to high ratings for Trump among self-identified Republicans in opinion polls. Even if we trust these polls — which should be taken with a grain of salt, given how poorly they predicted the election — this ignores independent voters, who remain about a third of the electorate in most states. Just because Trump has high poll ratings among self-identified Republi-can voters does not mean he is universally approved among independents who might vote for some Republicans. The very same polls say otherwise.

On the flip side, we are told that it is suspicious that more people voted for Biden than for Democrats down the ballot, and that “Biden-only ballots” are proof of fraud. But again, this is not anomalous. Voters who turn out only for presidential races frequently leave the rest of the ballot blank. In states with contested Senate races in 2016, there were 1.8 million more votes cast in the presidential race than in the Senate races. More than three times that many voters in 2016 left the ballot blank for House races. In Georgia in 2016, 248,220 voters — almost a quarter of a million people — cast ballots in the presidential race but not the Senate race. That’s 5.9 percent of all ballots. Leaving the bottom of the ballot empty was particularly pre-dictable in Pennsylvania, which for the first time in 2020 did not permit automatic straight-ticket voting (i.e., one mark to select the same party’s candidates for every office), a practice many Pennsylvania Democrats were accustomed to.

Poll workers tabulate ballots at the Allegheny County Election Warehouse in Pittsburgh, Pa., November 6, 2020.

An observer watches a poll worker tabulate ballots at the Allegheny County Election Warehouse in Pittsburgh, Pa, November 6, 2020.

Election officials process absentee ballots at a central count facility on Election Day in Kenosha, Wis., November 3, 2020.

Election officials canvass absentee ballots received on Election Day at a central count facility in Kenosha, Wis., November 3, 2020.

An employee of the Fulton County Board of Registration and Elections processes ballots in Atlanta, Ga., November 4, 2020.

Employees of the Fulton County Board of Registration and Elections process ballots in Atlanta, Ga., November 4, 2020.

Mail-in ballots are counted in Lehigh County, Pa., November 4, 2020.

Mail-in ballots are counted in Lehigh County, Pa., November 4, 2020.

Counting mail-in ballots in Chester County, Pa., November 4, 2020.

A poll challenger looks at absentee ballots to be counted on Election Day being processed by poll workers at the TCF Center in Detroit, Mich., November 2, 2020.

A poll worker gestures while processing absentee ballots at the Milwaukee Central Count the night of Election Day in Milwaukee, Wis., November 3, 2020.

Election workers sort through some of the thousands of mail-in ballots at the Orange County Registrar of Voters in Santa Ana, Calif., November 2, 2020.

More broadly, the voters who support Trump are not, in all cases, the same people as the voters who support other Republicans. You do not need a ton of familiarity with American politics to notice that there are still people out there who vote for Republicans at every level but are not fond of Donald Trump, and people who like Trump but are not straight-ticket Republican voters. When you drilled down to the county and precinct level in 2016, and again in 2020, you could see that in some places Trump did significantly better than Senate or House Republicans and in others significantly worse. Millions of people split their tickets in 2016. This is, historically speaking, normal.

The same goes for changes in the states and counties that are traditionally bell-wethers. As voting coalitions shift, so do the people and places that form the center of the electorate. Before 2020, only one candidate (Nixon in 1960) had won both Ohio and Florida and lost the election. New York and California were once crucial swing states. Mitt Romney won voters over 30 and a raft of other swing groups — independents, white Catholics, suburban women — but he lost because the center had shifted away from them. This happens all the time in American politics, and with a candidate as unorthodox as Trump, in a year of a once-in-a-century pandemic, it should not surprise us that it happened again.

Voter fraud and election fraud happen; the American election system is not fool-proof. In elections decided by a few dozen or even a few hundred votes, it can be a real concern. But in an election decided by 10,000 votes in Arizona, 12,000 in Georgia, and 20,000 in Wisconsin, let alone 81,000 in Pennsyl-vania and 154,000 in Michigan? In the age of computerized records and data analysis, fraud on that scale across multiple jurisdictions simul-taneously is mind-bogglingly diffi-cult to pull off without leaving significant footprints.

The turning point for the difficulty of concealing voter fraud on that scale was the 1982 Illinois governor’s race, which was almost stolen by Democrats by means of 100,000 fraudulent votes in Chicago. That scenario presented the ideal conditions for fraud: a single city with a well-oiled one-party machine that had many decades of experience in such shenanigans. It was still not truly that well hidden: A grand jury was shocked at the “boldness and cavalier attitude” of Chicago Democrats. With the aid of “a unique tool in the investigation of vote fraud, namely, the use of a computer,” the FBI obtained sufficient evidence to convict 63 people of fraud.

Courts have thrown out a few close races for fraud since then, but nothing on a similar scale has been observed. And in the absence of evidence, nothing in the results of the 2020 election should compel us to believe that something comparable happened this year.